Open AccessWeather-Aware Fiber-Wireless Traffic Prediction Using Graph Convolutional Networks
Author(s) -
Mariam Abdullah,
Jiayuan He,
Ke Wang
Publication year - 2022
Publication title -
ieee access
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.587
H-Index - 127
ISSN - 2169-3536
DOI - 10.1109/access.2022.3203420
Subject(s) - aerospace , bioengineering , communication, networking and broadcast technologies , components, circuits, devices and systems , computing and processing , engineered materials, dielectrics and plasmas , engineering profession , fields, waves and electromagnetics , general topics for engineers , geoscience , nuclear engineering , photonics and electrooptics , power, energy and industry applications , robotics and control systems , signal processing and analysis , transportation
In recent years, there has been an increased demand for better and faster communication networks to meet the needs of Beyond 5G network. The fiber-wireless integrated network has been widely investigated, taking advantage of the large capacity and low transmission loss properties of the optical fiber to extend the coverage of wireless networks and the cellular networks in particular. To optimize the resource allocation in these networks, machine learning (ML) techniques have been proposed, which aim to predict the cellular traffic in advance to allow proactive resource allocation. Existing works mainly consider two factors in traffic prediction: the spatial correlation amongst nearby based stations and the temporal dynamics observed in historical records. In this paper, we study a crucial and yet unexplored aspect, i.e., meteorological factors, such as rain, wind, or temperature. We first perform statistical analysis on a real-world dataset, the results of which confirm the strong impact of meteorological factors upon traffic volume. Thereafter, we propose a traffic prediction model that captures the temporal, spatial, and meteorological patterns simultaneously. The proposed model learns the network traffic patterns through a novel graph convolutional network - gated recurrent unit (GCN-GRU) cell. The GCN-GRU has a hierarchical structure with two child GRUs capturing the temporal dynamics in traffic and meteorological records respectively and a parent GRU capturing the unified impact of historical traffic and meteorological data upon future traffic. The spatial correlation among traffic records is captured by a graphical convolution module within the proposed GCN-GRU cell. We conduct extensive experiments on real-world datasets. The results confirm the effectiveness of the proposed model, with a performance improvement by up to 24.8% achieved using the hierarchical GCN-GRU model.